I S.. LMSC-D63 3363'1DECEMBER 1978
rI ~A FUNDAMENTAL STUDY
* OF FLOW AND FRACTUREI ~IN BERYLLIUM(
FINAL REPORT
ARO PROJECT P-13387-MS
I ~CONTRACT DAAG29-76-C-0051D CC.. DIT-RIBT'UnN ?TAThV!-FNT A !
DA Appostiuticj fh., ILAppr 'O, OT. f 1 ý. 1 1979
C..*LJU~~
ICHEL oe
L-OCKHAEAED
PALO ALTO RESEARCH. LABORATORY
LOfCKHEEDL MISSILES & SPACE COMPANY INC * A SUBSIDIARY OF L-OCKHE]ED CORPORATION
I,'A.,( AL-TC tLALIIOURNIA
I ~LMSC-D63 3363DECEMBER 1978
IA FUNDAMENTAL STUDYOF FLOW AND FRACTURE
I IN BERYLLIUM
( FINAL REPORT
ARO PROJECT P-13387-MS
I CONTRACT DAAG29-.76-C-0051
LOKHEPU T
~~~~PALO ALT F5ARHLBATO CLFRYI
Unciassifiled __________
SECURITY CLASSII`Ici.Ala OIF ThIS PAGE (miewi Voet Enteredj
REPORT DOCUMENTATION PAGE READ INSTRUCTIONS___________ BEFPORE COMPLETING FORM
1. REPORT NUMBER - 2. 50OVT ACCESSION No. 3. RECIPIENT'S CATALOG NUMBER
j\ 3387.2-MS. 1 ___________
-- T I L E (and Subtitle) ( Fna
-A Fundamental Study of Flow and Fracture --p:r,1~~~~ E
"t in Beryfliumo 1 Jun 76 - 31 Oct 78
7 G. CONTRACT OR GRANT NUMBE R(&)
Dona I dWebste r I AAG29-76- C-X051
9. PERFORMING ORGANIZATION NAME AND ADDRESS 10. PROGRAM ELEMENT. PROJECT. TASK
Lockheed Missiles & Space Company, Inc. AE OKUI UBR
Palo Alto, California
11. CONTROLLING OFFICE NAME AND ADDRESS EOTDT
U. S. Army Research Office r
P. 0. Box 12211 .- _ -- LResearch Triangle Park, 1C 2770953
14. 0WONIT"DRING AGrNCv NAME & ADDRESS(iI different froin Contro~liri Office; 15. SECURITY CLASS. (of this report)
~ ~ i L Unclassified-I15aS. DECL ASSI FICATioN/OWNRAOING
SCHEDULE
16. DISTRIBUTION STATEMENT (of this R~eport)
Approved for public release; distribution unlimited.
17. DISTRIBUTION STATEMENT (of the abstract entered in Block~ 20. if different from Report)
16. SUPPLEMENTARY NOTES
The view, opinions, and/or findings contained in this report Fire those of theauthor(s) and should not be construed as an official Department of the ArrnyPosition, policy, or decision, unless so designated by, ether documentation.
19. KEY WOPDS (Continue on reverse. side if necessary and Identify by block num~ber)
Beryllium Powder Grain Boundary FractureBeryllium Alloys Grain Boundary FlowBeryllium Recrystallization Grain Boundary Dislocations
IRecrystallization
Beryillum Ingots
20 ABSTRACT Conlin-, on *voevs, side It necssaery ldIdentify by block number) The ettect of thermomechaniclatreatments on t0e g rain size of HI 7b eryllium powder block and cast berylliumalloys has been studied. A grain-size controlle6-fractu~re mode change has beenobserved in both powder-source and ingot-source beryllium. The mechanism of recry z-allization In heav~ily deformed beryllium has been observed to be the in situ trarl!-formation of subgrains to grains by dislocation migration front grain interiors tograin boundaries. Beryllium which has been cold-worked and partly recrystallizedto form some very fine grains was found to Le much tougher than either unworkedmaterial or fully recrystallized matej_ and a proached the toughnes of hi h-
DDor IOR 1473 EDITION OF I NOV 5iS OBSOLETE &.~~
JAN 7a - . - . . . . . . . . -
unc !as si flea 13387. 2-MSS(CURITY CLASSIFICA'IONl OF THIS PAGE.(Oh.e Data nvrte33d)
20. ABSTRACT CONTINUED
strength a]uri num alloys. In upset-forged beryllium where the texture is similto that of hot-pressed block, strength, ducti l ity, and toughness were improvedby partial recrystallization. The enhancement of ductility and toughness arebelieved to be a result of grain-uoundary sliding in the regions where veryfine grains occur. Transmission electron microscopy has been used to show theearly stages of grain-boundary fracture and the grain-boundary dislocationstructure preceding fracture. This work indicates that the use of rapidlycooled beryllium powder rather than impact ground powder would allow a furtherincrease in ductility and toughness.
UncIass ifled ____
I.MSC-D633363
I
I SUMMARY
The effect of thermomechanical treatments on the grain size of HIP beryllium powder
block and cast beryllium alloys has been studied, A grain-size controlled-fracture
I mode change has been observed in both powder-source and ingot-source beryllium.
The mechanism of recrystallization in heavily deformed beryllium has been observed
to be the in situ transformation of subgrains to grains by dislocation migration from
grain interiors to grain boundaries. Beryllium which had been cold-worked and
I partly recrystallized to form some very fine grains was found to be much tougher
than either unworked material or fully recrystallized material and approached the
toughness of high-strength aluminum alloys. In upset-forged beryllium where the
texture is similar to that of hot-pressed block, strength, ductility, and toughness
were improved by partial recrystallization. The enhancement of ductility and tough-
ness are believed to be a result of grain-boundary sliding in the regions where very
fine grains occur. Transmission electron microscopy has been used to show the
I early stages of grain-boundary fracture and the grain-boundary dislocation structure
preceding fracture. This work indicates that the use of rapidly cooled beryllium
powder rather than impact ground powder would allow a further increase in ductility
and toughness.
I \
II ii. . .
LOCI,HEED PALO ALTO RESEARCH LABORATORY
aO40 Ff I D6 I & SFA CI C 0M PAN V I NC
A U I I D IA A Y O F 1 0(6C I0 AIIKI IA PA CAAPOIATION 1t
ILMSC-D633363
ICONTENTS
Section Page
SUMMARY ii
j ILLUSTRATIONS iv
1 OBJECTIVE 1
2 BACKGROUND
3 MATERIALS 3
4 EXPERIMENTAL TECHNIQUE 4
5 RESULTS AND DLSCUSSION 5
5.1 Grain Refinement By Recrystallization 5
3 5.2 Fracture Mode Transition 17
5.3 Microstructural Observations of the Initial Stagesof Fracture 28
5.4 Impact Resistance 39
5.5 Tensile Properties 46
1 5.6 Discussion 49
6 CONCLUSIONS 52
1 7 REFERENCES 53
II!!
II-tiii
SLMSC -1)633363
IS
i ILLUSTRATIONS
I Figure Page
1 Effect of Degree of Rolling Reduction on the Grain Size of High-Purity Cast Beryllium (EFI) Rolled With Intermediate Recrystal-lizating Anneals
2 Effect of Degree of Rolling Reduction on the Grain Size of CastIBeryllium Containing 0. 13% Cr Rolled With Intermediate Recrystal-lizating Anneals 7
3 Effect of Degree of Rolling Reduction on the Grain Size of High-Purity Cast Beryllium (EFI) Rolled Without intermediate Re-cry-stallizing Anneals 9
4 Effect of Degree of Rolling Reduction on the Grain Size of CastBeryllium Containing 0. 1970 Ti Rolled Without IntermediateRecrystallization Anneals 10
5 Effect of Annealing Time and Temperature on the Grain Sizeof High-Purity Cast Beryllium Rolled 93'Y(, After Upset Forging 11
6 Effect of Annealing Time and Temperature on the Grain Size ofCast Beryllium 0. 13% Cr 12
7 Isothermal Grain Growth at 977 K for E I and Be 0. 13% CrAlloys 13
3 8 Optical Micrograph of EFI, Rolled 9670 and Annealed 977 K5 Minutes 14
9 Optical Micrograph of Be 0. 130 Cr Roiled 961(, and Annealed977 K, 5 Minutes 15
10 Transmission Electron Micr(graph of E F, Rolled 961", andAnnealed 977 K, 1 Minute 16
11 Transmisvion Electron Micrograph of EF1 Rolled 96/'(, andAnnealed 977 K, 1 Minute 18
12 Transmission Electron Micrograph of Be 0. 13% Cr Rolled90% and Annealed at 977 K, 5 Minutes i9
13 Scanning Electron Micrograph ol EFI Rolled 96',,, Annealed977 K, 5 Minutes, and Fractured at Room Temperature 21
14 Replica of Fracture Surface of Cast Be 0. 19'ý,, Ti, Reduced 907,"(by Rolling and Partially Recrystallized at 977 K for 5 Minutes 22
I iv
IL OCKHEED PALO ALTO RESEAPCH LAF3ORATORYt 0 r K " a 1 0 M I % S I t L I S A P & A I C (O MA h Y I N C
A S Q D IS1 I1 P LOCKINI10 AIC IA PI (ORPOaAi ION
I I.MSC -1)(;33363
!I IPage
15 Replica of Fracture Sut-face of HIP 1. 56% BeO (1111243)Extruded 10:1 at 1033 K and Annealed for 1 hr at 1144 K 23
16 Replica of Fracture Surface of HIP Be, 0. 70 BeO, Upset-Forged 1033 K, Reduced 75/1 by Rolling at 922 K, andAnnealed at 977 K for 10 Minutes 24
1 17 Higher Magnification View of Sample in Fig. 16 Showing a
Region of Fine Intergranular Fra( ture 25
18 Grain-Size Controlled Fracture Mode Transition for High-Purity Ingot Source Beryllium (EFI), Upset-Forged andAnnealed at Either 1061 K or 1116 K 26
19 Grain-Size Controlled Fracture Mode Transition for High-Purity Ingot Source Beryllium (EFI), Warm-Rolled 96/,and Annealed at Various Temperatures 27
20 The Effect of Temperature on the Grain-Size ControlledFracture Mode Transition for High-Purity Upset-Forged,Ingot Source Beryllium (EFI) 29
121 Grain-Size Controlled Fracture Mode Transition for High-Purity HIP Power Block (BOP 32) After Upset-Formingand Warm-Rolling 30
22 Grain-Size Controlled Fracture Mode Transition for High-
Purity HIP Beryllium RR243 After Extrusion and Annealingat Various Temperatures 31
: 23 Grain-Size Controlled Fracture Mode Transition for High-Purity "As-HIP" Beryllium Block (BOP 12 and BOP 22) 32
24 Transmission Electron Micrograph of Cast Be 0. 10% Ti,Warm-Rolled 96%, Annealed 977 K 10 Minutes, and ThenCold-Rolled 15% -3
25 Transmission Electron Micrograph of High-Purity Pow&drSource Beryllium Alloy BO01 32, Upset-Forged, Rolled 75%,and Annealed 971 K 10 Minutes 35
26 Transmission Electron Micrograph of Be 0. 7% BeO (BOP 32),Upset-Forged, Rolled 75% at 922 K, Partially Recrystallzcdfor 1 hr at 977 K, and Reduced 10% by Cold-Rolling :36
27 High-Voltage (650 kV) Electron Transmission Micrographof a ReLatively Thick Foil of 1301) 32, 0.77 BeO, Upset-Forged and Rolled 75-7 at 922 K, Partially Recrystallized at977 K for 5 Minutes, and Then lieduced 10" at RoomTemperature
i v
LOCKHEED PALO ALTO RESEARCH LABORATORY[ 0 t 1 0 " $ I % I t % A t -I ( O MI A . I INC
I ¶ a ýI • , A 1 0d 0 L I 0 1 t A A, I P i A I I D N
-
LMSC -D633363
aI Figure Page
28 Transmission Electron Micrograph of BOP 32, 0. 7/(, BeO,Upset-Forged and Rolled 75:N. at 922 K, Partially Recrystal-lized 977 K 1 hr, and Deformed by Rolling 10"', at RoomT e mpe rature 38
29 Transmission Electron Micrograph of EFI Polled 96K/ý,Annealed 977 K I hr, and Cold-Rolled 14K, 40
30 Impaict Toughness of Ingot Source Pe 0>, 5K:•/ V, After VariousThermomechanical Treatments 41
31 Impact Toughness of High-Purity HIP Beryllium AfterExtrusion and Annealing 42
:32 Impact Toughness of High-Purity HIP Beryllium With TwoLevels of BeO After Extrusion and Annealing 43
33 Impact Toughness of High-Purity HIP Beryllium With 'IwoLevuls of BeO After Extension and Annealing 44
34 Impact Toughness of High-Purity HIP Beryllium Block(130P 32) After Upset-Forging and Warm-Rolling 45
:35 Impact Transition Temperature for High-Purity HIP Beryl-lium (BOP 18) Extruded and Partially Recrystallized 471 36 The Effect of Grain Size on the Charpy Impact Energy of aVariety of Partially Recrystallized Beryllium Products 48
'37 The Effect of Annealing Temperature on the Strength,
I Ductility, and Toughness of Upset-Forged B~eryllium 51
SI
IIII
vi
L-OCKHE EL) PALO I-TO RESEARCH LABORATORV'AO (I o Iv I 1 0 A I I C & A T | 0 0 F 0 A I 1 0N k
A St bSU II * *A O • OCS MI I O I *I CSIAS C ¢O3P@O SA TO N
ILMSC-D633363
f1J Section 1
OBJECTIVE
The objective of this program was to develop an imderstanding of the fundamental flowarid fracture mechanism in beryllium. A specific objective of this program was to
indicate how new mechanisms of flow and fracture could be utilized to overcome beryl-
lium's low toughness.
II
If
LOCKHEED FPALO ALTO RESEARCH LABORATORYSO kk I C I D 0 1 I I•1 1 S $ t - A ?C4 f •MO A 4 1
LMSC-D533363
I
Section 2
BAC KGROUND
Although modern beryllium has the highest structural efficiency of any material or
composite, its use has been restricted by low ductility and toughness. In the last
5 years, the ductility of beryllium has been dramatically increased so that 6-percent
elongation in all directions of a semicommercial, hot isostatically pressed, low oxide
(0. 57o) block can be produced routinely (Ref. 1). Still higher values of 13-percent
elongation in virtually nontextured upset forgings of the same material have been pro-
duced experimentally (Ref. 2) and recent Russian work (Ref. 3) using upset forging
of high-purity cast material has enabled three-dimensional tensile elongation of
22 percent to be obtained. For most structures, this degree of tensile elongation is
more than adequate to permit designs that can utilize the full strength of the beryllium.
g At this stage, beryllium compares favorably in its mechanical properties with other
engineering alloys in all aspects other than its resistance to impact shocks in the
presence of a stress c oncentration. This condition is simulated experimentally by
charpy impact testing which is used on this program to evaluate the effects of micro-
structural and mechanistic changes on toughness.I
I
LOCKHE ED PALO ALTO RESEARCH LABORATORY
A o t IIb A I No t A ( lt A oa I o I I o
- -
LMSC -D633363
Section 3
MATERIALS
Two basic types of beryllium were used on this program:
(1) 11i-b-purity powder products formed from electrolytic flake and then hot
isostatically pressed (HIP) by Kaweeki ferylco Industries (KBI3)
(2) Vacuum-cast ingots 2- to 3-in. -diameter containing small (0. 1 -0. 5"")
addition ol titanium, vanadium, or chromium plus a high-pur-ty vacuum-
cast ingot (EFI) containing no alloying additions
Table 1
CILEMICA L COMPOSrTION OF MATERIA .,S
__C"i"IpositioflP I1M
ABloy Form Ie() Mg Si A) F .1 Ni .C' r V
F I Ingot 40 10 100 9 0- 29), N. !). N. ). N. 1). N. L).
B(e- lri Ingo! N. I). 2() 380 67 o 5 !).2 1900 1:10 40 80
Be-Cr Ingot N. 1). 10 10 o 0 ,tk 714 N. La. N. I. 1:300 N. I).
be-V Ingot N. 1). 10 400 6(0 G :( 0 1,5s o 17o :31, 5700
H P)1' 11 1Ji1) B•ock 5, 1o0 A6 90 1 35 14(6 !220 N. i). 1i5 N. 1). N. I).
1()H) 12 1fill) Block 4, ýOO 2 6 9 ( 3" H 6 22o N.- D :95 N. I). N. 1).
HO P 16 HIlP Bllock 5. 01)0 50 110 35 140 1i(o N. !i. .11 N. I). N. 1).
Io)1) 18 lloi lock 5,100 32 150 30 250 3'7;- N. 1). 26G N. 1). N. I).
1b()P "'2 1111) llh)ck 5, :o) 37 70 381 180 1SO0 N. 1). '160 N. 1). N. I).
HIP':i() 1111 f ilhck 6,7o0 41 120 35 i 280 N. 1). 1_65 N. )'. N. 1).
I ,i )) 32 fill) Block 7, mm 1 45 33 if)21,I(1 N 1). 5) N.1). N.1).
It 2I t ii3) H l hI 'k , ,(' ) :G o J 16 5 ) , N.5 t) 2 0IN,1).
N. D. Not )D1 ti rni itid.
Luk,)1,HF ,- D PAL.) ,ol T') N2 10 ARCH L ABORA!.ORY
Lot~ ADC I ,i Ii I 1A~ 0~~ 4 h1
-LMSC-D)633t36:3
Section 4
EXPERIMENT 1, TECHNIQUE
All beryllium samples were canned in mild steel prior to forging extrusion or rolling
to prevent surface cracking. Upset foring was carried out at 1)33 K (1400°1F), rolling
at either 1033 K or 922 K (1200'F) and extrusion (10:1 reduction) at 894 K (1150' F).
Fractographic examination was performed by scanning electron microscopy for coarbo'
grained (> 50 gm) samples and by transmission electron microscopy of shadowed
replicas for finer-grained materials.
Direct transmission electron microscopy of the beryllium was conducted after electro-
polishing in a solution containing 82% ethylene glycol, 5%1 H20, 9% HNO 3 , 2% t 2 So 4 ,
and 2!,,V) HCL. Both conventional (1.20 kV) and high-voltage (650 kV) electron microscopes
were utilized to examine the electropolished foils.
Charpy specimens for impact testing were made with the rounded notch, 1-mm radius
and 1. 5-mm deep conventionally used with beryllium. The surface was etched before
testing to remove 0, 1 mm of potentially damaged surface material. The specimens
were tested on a 24-ft/!b impact machine.
Upset forging was conducted as described previously (Ref. 2).
4
"LOCKHAEED PALO ALTO RESEARCH LA0OIATCRY
I n C':-'•" " .0 ". .0* .. 6 I ta i C0Ml PA h I1 . f -r
A S U I . 1Ai A OP t a( 1,. I 0 A '~I C AF I COPPOEV,00 .0N
LMS(,-I)6.:3,3(;,3
Section 5
RESULTS AND DISCUSSION
-5.1 GRAIN REFINEMENT BY RECRYSTALLIZATION
Grain refinement in beryllium, as in most other metals, confers property improve-
ments such as higher toughness and tensile elongation (Ref. 1) together with higher
strength (Ref. 4), but there are limitations to how far this refinement can be taken.
In powder block beryllium, the grain size is similar to that of the input powder and
any degree of grain refinement can be obtained by reducing the powder size. However,
this increases the surface area of the powder and hence the oxide level. This counter-
acts the benefits of grain refinement on toughness and ductility, and optimum properties
are observed when the grain size is approximately 10 /m. In cast beryllium where
oxide is not normally present in sufficient quantities to act as a grain refiner, it is
very difficult to produce a recrystallized grain size less than 30 pm even in sheet.
An alternative method of grain refinement investigated in this program is to use inter-
metallic compounds of titanium, vanadium, and chromium to act as grain refirners in
cast and wrought beryllium.
Two types of thermomechanical treatment were conducted to determine which was
most effective in producing grain refinement. In the first type, the beryllium was de-
formed about 20 percent and fully recrystallized at the minimum temperature at which
this could be accomplished. The recrystallized material was then rolled to about
40 percent and recrystallized again. This process was repeated for 90- and 96-percent
reductions. The grain size was reduced ateach step as shown in Fig. I for high-
purity beryllium (EFI) and Fig. 2 for the 0. 19-percent Ti alloy.
The initial "ae-cast" grain size was taken as the dendrite width which is about ten
times smaller than the average dendrite length in both cases.
5
SLOCKHEED PALO ALTO RESEARCH. LABORATORYL 0 K I 1 I I 1 S I $ a I a I P A C f C0M• MPA h I INC
____A I USI IDOI Aft{Y 0 7 (IOC£WISO A ICI I A P C OIPOSAI' OI 0N
I lIvlI)(CI 6J6
III I-
DENDRITE WID1H HIGH-PURITY CAST BERYLLIUM (EFI)AS .-CAST ROLLED WITH RECRYSTALLIZATION
1000L AT THE STAGES INDICATED
.... 1255 K 30 MIN •E
I-.
"Uj I HRU 0S1172 Kj 01255 K 1 HR
I --Z_ 2 HR i
0100-I 1144 K I HRZ 01089 K 1 HR-Jjj 103KIH
911 K20 HR
10 I I I I _0 20 40 60 8o 100
PERCENT REDUCTION
Fig. 1 Effect of Degree of Rolling Reduction on the Grain Size of High-. PurityCast Beryllium (EFI) Roiled With Intermediate Recrystallization Anneals
6
LOCKHEED PALO ALTO RESEARCH LABORATORY•LOIO 1 1., 1 *.1r I 1 Y AC C 0 M PANY A N.
A •ItGOIAtY O P LUC*KIEO A IC AII C 0 I P*A ION
I ~LMSC.-D63;3363
!
1000CAST BERYLLIUM CONTAINING 0. 19% TiROLLED WITH RECRYSTALLIZATION AT THESTAGES INDICATED
DENDRITEE WIDTH
AS-CASTANNEALED
uu1255 K, 30 MINI-. 1089 K
z 100 1 HR 0S~1225 K
1 HRI 01255 K I HR-- . 1089 K I H R- 81033 K I HIR
:1 zujL
911 K 20 HR
10- I0 20 40 60 80 100
PERCENT REDUCTION
Fig. 2 Effect of Degree of Rollintg Redtuction! •rn the Grain Size of Cast BerlliunContaining 0. 13;,, Cr Rolled With Intermediate tlecrystallizing Anneals
7
LOCKHEED PALO ALTO RESEARCH LABORATORY
{.EID1 MrP IC I CP A I CO#PA NV I NC
'A S U 5 S I1D I A I T 0 0 LOCk I0N1D 6 A II ( I A•P T C O P9O 0 ATION
~ Ae*' A4 ~SPT~~~'~l~# ~~P n
.•m,•,J nmmmamm emim•i
LMSC-D633363
•. the second type of thermomechanical process, the specimens were rolled 90 or
96 percent before recrystallization anneals were performed. These results are shown
in Figs. 3 and 4 for EFI and 0.19•Xj Ti, respectively. A comparison of the two •srpes of
thermomechanical treatments shows that the direct reduction wi•.out interstage anneals
is more effective for both materials in producing grain refinement. This result follows
from an examination of the recrystal]ization mechanism as will be discussed below.
The 0.19% Ti addition can be seen to produce grain •'efinement in both processes and
also produces some refinement of the as-cast dendrite size.
Recrysmllized grain sizes after upset forging and rolling are shown in Figs. 5 and 6 for
EFI and a 0. 137/• Cr alloy. There appears to be no particular advantage to any particu-
lar temperature in terms of grain size, but annealing temperatures above 950 K pro-
duce fully recrystallized structures in less than 72 hr.
A more detailed study of the recrystallization of EFI and Be-0.13% Cr was conducted
at 977 K (Fig. 7). The recrystallized grain size was measured in both alloys in the
early stages of recrystallization when the structure was only partially recrysmllized.
Recrystallization is obviously retarded in the chromium alloy, probably by the in•r-
metal lic compounds.
In this ahoy, grain growth is continuing in the recrystallized areas even though further
nucleation of recrystallizeo grains is inhibited. The optical microstructure ol EFI and
Be 0.13'/•, Cr after 5 rain at 977 K is shown in Figs. 8 and 9, respectively. In both
mamrials faint, ghostly :'grains" can be seen inside what appears to be large unrecrystal-
lized areas. These are actually subgrains in the process of transforming to grains.
This process As revealed in more detail by transmission electron microscopy.
The subgrain structure of EFI in the vicinity of an origimal grain boundary after roll•g
95 percent and annealing for 1 min at 977 K is shown in Fig. 10. Most of the subgrains
have poorly defined boundarie• and a high density of dislocations in the subgrain interiors.
A few of the subgraL-m, such as those at A on the original grain boundary, have low
•i LOCKHEED F'ALO ALTO RESEA•'CH LABORATORY
L O € • N I I • Jo•F| $ILIFI & • P A ( | € 0 M • A N Y , IN (
A $',• S | I O• A I Y OP L O(I(•IID A I II C I1" P Ir ( 0 IpO, A ( IG N
.. ., ., •, .. • !
LMSC -1)633363
DENDRITE WIDTHAS -CASTr
1000 HIGH PURITY BERYLLIUM (EFI),ROLLED DIRECTLY TO THE INDICATEDREDUCTION BEFORE RECRYSTALLIZATION
E
:L
UjLUI-Zi1O0Z 1033 Ks
I HR
91K 16HFR
I you0K 16 HRz ~939 K
2OH ~ 950 K I HR
922 K j1. 5 HR
922 K I HIR10
0 20 40 60 80 100PERCENT REDUCTION
Fig. 3 Effect of Degree of Rolling Reduction on the Grain Size of High-PurityI ~Cast Beryllium (E FI) Rolled Without Intermediate Recrystall izinigAnneals
9
LOCKHEED PALO ALTO RESEARCH LABORATORY1 OC KN I 1 III t I i A IP A CI ( .p AN v N C
1 t~ bAi A O0F L 0 CK NII1O A I I(BIAP I C 0 2P 01AI 1 0N
I I.,M,;C -1)633:306
I
1 ~1000 -
CAST BERYLLIUM CONTAINING 0. 19% Ti,ROLLED DIRECTLY TO THE INDICATEDREDUCTION BEFORE RECRYSTALLIZATIONI
DENDRITE WIDTHAS -CAST
E
LUI LUI,- 100
Z 1033K IHR3. 939 K 20 HR
I 951 K16 HRS922K 1.5 HR
0950K IHR
1 922K IHR5 •I I I I I
0 20 40 60 80 100PERCENT REDUCTION
Fig. 4 Effect of Degree of Rolling Reduction on the Grain Size of Cast BerylliumContaining 0. 191 T1 Rolled Without Intermediate Recrystallizing Anneals
t 10
LOCKHEED PALO ALTO RESEARCH LABORATORY
M I IOC1 I L4 I S & S P A C & C 0 m PANT •N
A 9SU 1I I AlY Of L GCK I1 0 AI AICI AA f I CO POIA t1ON 0f
i--
7112
I--
0 0
E EU- :L IýU
0 W0
5,- UU
o/ 0Uc- LU
ce z
U- <J
EiZ
D <0
044
(s) I Owi NIV3NNV 44
LOCKHEED PALO ALTO RESEARCH LAE30RA-FORYi Ock NIot MI I st S PAI C QMPA N, N c
A UU6U1rASY OF 1OCK4111) AIPCOAPI CO0IPO# ftA TON0
LMSC-D633363
%0
04 C1u-i
z 0 0
0<0Z E)
E C
00UU)
t<z /
E "F .
> -- UJw 0.L
kn-
00)
0 "u -I U)
j N N
44
LOCKHEED PALO ALTO RE.SEARCIA LABORATORY
LO I 0 II ( a . I 1 0 m I I t I S P A C II ( 0 M 01 A N y I m c
A S U I & I D I A 0 Y 0 FI t 0 ( KI[ 1 0 A 1 0 C• I A P T C 0 I P 0 0 A T 1I0NI
LMSC -D633363
010
-j'
ZLU* Ul
z 01:L oeo
z 00
10 LLuI 01C) I
0 z ý4
-J a zOL - Lu
O j LUJ -J0 IU-
0
LLLu
- J ad
do LA-..-)
oý1 uu I-
L@L~
0w' 0dIV3N NV0 0VNl 0V
r .13
LOCKHý-E PALO ALTO RESEARCH LIABORATORYPC f A ' £P* (O0MP A NY ~ C
A $U $ I BY Oý to (w k D A 'It A PY CC) RONA I O N
L.MSC -1)6:336:3
Fig. S O.pticail M\icr(igraijih of I.A., RolledC( andtl~ Amicaed 9'17 K, 5 MillitctS. VaintfgI1()stlýy suhgraill'ý idl Lw ht.'cfl in1 soflic lltni cC I\ S tlI1/( ( ar e~ls. Matgnification
[)UPAL,() AL-C RO , ARCH L-AEORATORY0 l 1 H IA I P N N
*~ ~~~~~~~~~ A,*~'A y O ~ ,*~m g ! ( ~ f~tO
LMSC -D633363
Fig. 9 (,pia AhcrMi r jpi ot he0 13,; CK Rlt~Ie 116 .ýid Amiii;il~t(I ýit 97 7 K,
Lo KH F D .AL A I, F'it ,f A NbPA'L A H,)R A] OW
I -.-- I-I--- i.I.--,*,-,-,-I - .. I . <-", ~ * 1 1- --
I, it I
big.~~~~~~~~~~ IC I t~i~ivi AKt I\.i t I i{IU V? IrtId 'N
LMSC-D633363
dislocation densities and may appear in polarized light to be the ghostly recrystallized
grains described above for Figs. 8 and 9. In other areas of the same sample, the re-
crystallization process has progressed furthe:" as shown for grain A in Fig. 11 where
the grain interior is dislocation-free and the boundaries are sharply defined. This
grain would appear in polarized light to be fully recrystallized. A still later stage of
recrystallization is shown in Fig. 12 which shows the Be 0. 13% Cr alloy after 90-percent
reduction ant an anneal of 6 min at 977 K. This area shows a boundary between unre-
crystallized subgrains and fully recrystallized grains with low dislocation densities and
high-angle boundaries, The particle at A is a beryllium-chromium intermetallic with
an associated defc;rmation void. With continued annealing, the unrecrystallized sub-
grains in all the cast alloys are seen to transform gradually to recrystallized grains of
tho same size, although there is no definite point at which a grain can be said to be re.-
crystallized by microscopic observation. However, there is a sharp transition in frac-
ture mode during annealing wlt`h can be used to define the point at which the conversion
from subgrain to grain is complete.
It follows from the above observation indicating the mechanisn) of rec rystallization to
be by subgrain transformation that the finest recrystallized grain size will result ro)m
the finest sul)grain size. ['his sutlgrain size is found in most metals to be inversely
I)ffo)()rtiOnal to the amount of cold work. This explain3 why in the two thermo)mechanical
processes described above the finest grain size was pr)duced when internmediate re-
,.rystallization stel)s, which woUld have r'educed the total am,)wlt of Cold work, were
aV()ide(d. If the recryVstallization mechanism had been Ib)y the gr(,wth ,,t gra n-I)ounda ry
nuclei, then the gradual grain refinenlent pr(duc lic ey the inte rmeld iate recry statliization
St lc) iln the second process would have produced the finer grain si:,e
5. ! ,{AC'1't I,: M DI)I' TRANSITI()N
The norimal fracture mode in I)oth I pOwer-souree and ingot-source heryllium is cleavage.
Si)w 'vr'(i, whenll Ohe g r;iin si. it i '(1ceýd IRe hA W a ce('v iitn critical value characteristic ()f
lic'h c(Ollp()siti( )l and IllethO(i 1)f l)anllUfacCtrll'e, thte frla:cture im alle, ) iH bY grain-b)(iinda 'y
,•|ar':tt .n. llPrcvio s wo)rk (Het. I!) has in(dicated grain-houndary sliding (#ccu[s [itohr to
17
L U(CPK F,4 E D PAL O ALTO RE sEARIISRC LA HOR IT )RY
, Cý . .U I ID I I & % & t P I S 6 .I .I 0| C
A s~.~~Or I O L .) 1 1 1 P 1** f O OS I CJ
I A I i I.* I t
A~~ M e~K AI~ 1, N
I MSC, -1)(833:361;3
r,1 2 1 Trai fl-t1is ;iomF l tl'I Mi ur i ( )i Iof H(1. 13',V (, 17 Rolled 90",' an[d A nnealed
til 977 iK, 5 Nll!, zI'n11;lC81iol 16,0O('00
L(,)( KHý ,UlI) F<At 0 ALI(, I' Fd APCH L ABORATORY'
I. ho t Itt 1 0 1 PM N 1,1 IJ
LMSC -D633363
IJ fracture, and this additional slip mechanism may be responsible for the improved tough-
ness and ductility of the fine-grained materials. The two fracture modes can be seen in
Fig. 13 which shows the fracture surface of EFI after rolling 96 percent and partially
recrystallizing for 5 min at 977 K. The unrecrystallized area, t,,pified by A, shows
cleavage with only minor perturbations in the fracture path across subgrain boundaries.
The recrystallized areas, such as B, have grains which are about the same size as the
subgrains but where fracture is completely intercrystalline. Similar fracture mode
transitions are observed in partly recrystallized powder products (Figs. 14-17),but the
transition is at finer-grain sizes than for cast and wrought products.
In Fig. 14, an isolated grain in a Be-O 19'Y% Ti alloy has fractured in an intergranular
manner while the surrounding grain fracture by cleavage. Figure 15 shows a mixture
of intergranular fracture and cleavage :racture in a HIP powder product. BeO particles
can be seen in the grain-boundary fracture surfaces. It should be noted that these frac-
ture characteristics are easily seen in powder products using transmission electron
microscopy of carbon replicas as shown in Figs. 14 through 17, but are very difficult
j to resolve with the lower contrast and resolution of the more commonly used scanning
electron microscope. Figure 16 shows the fracture of a high-purity HIP powder product
aifter upset forging, rolling, and partly recrystallizing. The fracture is mainly cleavage,
but some areas of very fine intergranular fracture can be seen at A. Figure 17 shows
an intergranular fracture area in this material at higher magnification
The fracture mode transition temperature for high-purity ingot-source beryllium (EFI)
after upset forging and annealing at two difierent temperatures i3 shown in Fig. 18.
All grains under 35-pm diameter fract'ire intergranulariy. The largest grahu sLze that
fractures along the grain boundaries varies from 100 pin in the ;ample annealed at 1061 K
to 400 pm in the sample annealed at 1116 K. This indicates that at the lowar annealing
temperature some excess dislocations remain inside the larger grains and encourage
cleavage fracture. A ulmilar fracture mode transition Is seen when EFI, which hac been
reduced 96 percent by warm rolling, is annealod at 997 K, 1089 K, and 1172 K (Fig. 19).
The main difference between the fracture of the relatively untextured upset-forged
material and the highly textured rolled materiel Is that the grain size at which all frac-
ture is lntergranular is displaced from 25 pm to 4 pm by the textur"e.
20
) LOCKHEED PALO ALTO RESEARCH LABORATORY
A m D I A a Y 0 f 1 0 C N M I I A I N k F IU 0 I F 0 I A IP 1 0 N
'ig. 13, Suamniln-, 1EIo(ltriO Mic-lograph)I fF] El iollc 9rv , Annettl1d 977 14 5 Mifli~keM
t (.)KHf It) P-AL() ALT M( tz-ýi ARCH L.AROPAY CRY
A ~~tUASI O(~~IO P'P INN"~
Ii.14 R~j)N!wa ý)I FFI IIcUf Sut nace () C',ast He o. 19 Ti, Reduced 90 y RýuIllngand Partially Recr Vý,allizcd at 977 K 1w, 5 111mi. All isulated 91rain -pdiairet44ýr hall fract ure-,d along its W)undary while thec surrom)nding gmai1 ns orsubgraiut s I hive failed by transgr~anlaUI cleavage. Magnuification 24, 000
I ()C HE f~ I) PAI(. At TO FO, '-. ARCH L. AfiORATJ')RYA I I 1 0PA 1 C 4A N I N
1 11~ XI f a ~ II, A * b * ( 3 P~ C
I'ig. I ( ~j~e )I 1týWAuIt, surl~tcc ()I IIIP I Bei1( ) I{ Lxtrudt(1 I(J:1 at Iul::i 1Km1id IAlnllaiCe tor' 1 fit Mt 1 144 K, TaIRI'atr SlhM s ainXtIC ()I IIUc (I -:i j111)
iftl1a~LitI IaiILuIIt lIIJ d~ tI*:IIS(,"*LI~)l' Ci~V1C INHIuC. H0u ) P:LrtiCleM
IýOCKHE F ID PAL G, A,_T( RE SI ARCH I AMM)A I RY
I DS A1 I AP I , C.
Fig. 16 liepI icaI (A Fractur Sc urface of IlP Be, W) 70 Beu), [Ipse V lu iged 03,3 K, lHe-(IUCfJ~l 75 ,by Rolling at 9J22? K :md~ AnneHaledJ at 977 K fur 1(1 min. TJhe striqucturemuainIl cle!avage ta ilu ye with i5 uhited a1'eiS !in jute 1gI'anulL r I racturMagnificat io H 000)
24LOCKHE ED PALO ALTO RFSFARCH'- L-AHURA7ORY
*~~ ~U CO A ~ C CN D AISCA o ( ,I ! o I
F ig. 1 7 H ighe r Magnificat~ion View of Sample in Fig. 16( Showing a Region of Fine Inter-grannlai Vracltur,Ž. ke( p~articles are \'isihlc 'mu Ow~ graiI1-ikoflind~tln SUIAWC.
L 0( K H f ) PAL 0 A L. T) RI1 5 FA F CH L A H (RA T R Y
L (k ,1t F A #A, 1P . 1
A~~ F1 1CS A'* 1(~tr 0A* i . . N
I LMSC -D633363
LUC4
LUC'
LIJLZ
6-E
ZLI N
00
LUJ
-Jc
00
0 0 10 0 040
26
LOCKHEED PALO ALTO RESEARCH LABORATORYk 0c N mI~ 1 ) 0 A IF Ac I cO0m pA~ 4 1
A tUS4ID'Al Of toCKI'E , AIRCIAlf CORPONAY''N
LS-D6,'6
j 0
%owo
Z rin
0- Z -eOwj
-j-
Q Q 04
uJJ 0JU
8000 0 Lo
2 7
LO)CKHEED PALO ALTO RESEARCH LABORATORYk I a0 mI' IOC NL ~ I I a I (PA Ct C 0M FA N I (
A )~I S1 01A( 4 O0I LO0 (9 N 6 A 1 0C III A OIT C 10 bA IO10N
LMSC-D633363
Ij It was expected that an increase in the temperature of fracture above room temperature
would significantly increase the tendency of the beryllium to fracture in an intergranular
manner and also increase the grain size at which intergranular fracture occurred. A
small increase in the latter parameter was observed in EFI (Fig. 20); however, in powder-
source materials which have only about 10-percent intergranular fracture after cold work
and recrystallization, no significant increase in the amount of intergranular fracture was
observed up to 533 K, although the observed intergranular fracture became better defined
at the higher temperature.
Fracture mode transitions for high-purity, HIP, powder-source beryllium are shown in
Figs. 21-23. These materials show comp.etelv intergranular fracture for grain sizes
below about 1-pm diameter. Intergranular fracture at the larger grain sizes seems to
be only in areas where little or no oxide particles occur, i.e., when rccrystallized
grains form inside the original beryllium powder particle bounlaries. This is particu-
larly true in the case of unworked HIP material where the amount of intergranular frac-
ture is negligible and is seen only when multiple grains form inside a single powder
particle and do not have oxide particles along their boundaries. These isolated regions,
therefore, behave like ingot source and favor grain-boundary fracture.
5.3 MICROSTRUCTURAL OBSERVATIO)NS OF THE INITIAL STAGES OF FRACTURE
The above results have shown that fine-grain materials fracture prcidominirntly along
the grain boundaries. Previous work (Ref 2) on polished samples has indicated that
grain-boundary sliding precedes and is probably responsible for intergranular fracture.
However, to examine the inicrostructural events leading to grain-.botudary fracture,
recrystallized and partly recrystallJzed samples were deformed by cold-rolling at
room temperature and examined by transmission electron microscopy. Specimens
were examined ir, the areas containing the finest recrystallized grains for evidence of
grain-boundary sliding or local deformation in boundary regions that could Lead to
grain-boundary fracture. The type of structure formed in a partly recrystallized ingot-source Be 0. 191, Ti alloy after 15-perceii, reduction by rofling at room temperature
is shown in Fig. 24.
r 28
LOCKHEED PALO ALTO FRESEARCH LABORATORY
SO PN Ih A { (I CO M A h I
A 6 S 1 0 1 A A V OP F CL 0 C K # 1 0 A II C I A , T CO kO P 0 T 1 0 N
LMSC -163:3363
040
a 00
u.4
LiLU
Ix U-
UU-
~Zoe I000
0LLJ-1 cd
< U- ;-4
< 0, 0)0fl '0
J ]~~~~~fL.L- V VfNV~1
LA29
LOCKEED ALO LTORESERCH ABORTOR
O(Mw ED M I1IL~ U PAC* t0'A wy N
IUS1~rA~ OP LO( NI~r A~(UAY Co'OSA ~ C.
I LMSC-D633363
4-
0Lii
(5 01
OZO
CcC
M
-4,
0
- $-4
0O
U'
Q,
040
30
LOCK~HEED PAL-O ALTO RESEFARCH LABCJHATORYk OkN.1 IC, M. * ,I IFA~I. co,% AN' I C
A I U iSI 1A* 1 O0 LO .. AIID A~I(IAPV CoppoI.alIok
LMSC-D)63 3363
41004
ol
Q -4
C-) 0
00IuIIE'0
C. a)
0~L Q) 00
0 Cl
LOCKHEI-D PALO Ai..Tt- RU,E~ARCH L ABORAT ORYto( k 01 1t - St4,I' lt A vP C I ,. 0 1.N t
%U8110 - 21, Op .111*P * B I I ONIONO 10* f,
I ~LMSC -D6 33363
I -
04 (%4
4.)
PU-.
* -4
Li
0 0 0 ,.;C8
324
LOCKHEED ~ ~ ~ ~ ~ C LAOATORSARHL6OAO)~~~~~~~~- MI IgS rCI ( M AN
A I~i~~AI~ P L C~4IO AI(BA ! C 4POR ICO
Fi g. 24 PIr:ifnlxiiisionf J'Jectron Mlic rogi iph of Cast lie 0. 1~ P1'i, '.11 Wrv iRollcd 9Arawaled 977 K .10 mnin,aiid T he ii f Ad Blled 151ý,. A smai ll cr < hts~ ap--pearu on( UfLeU gra~in b)oundriQy ia. A, Dis IoCation. tanglei ýit, triple( junlction I
hei e'1 v(idence tot slidinug aiung, one ot icLl adj ccat. hound.'IriO.eMit~gni.ficatin 2l12, 000 )
LOCKHEED PALO ALTO RESEARCH L-ABCORATQ,)Y
w 0 p ~ t j) A *( I t C O T
LMSC-D633363
Beforte deformation, this area contained very fine (- 3 pmo) recrystallized grains of low-
dislocation density. Deformation has produced a fairly uniform network of dislocations
throughout the grains. A small (1-pm-long) crack has Lormed at a triple junction A and
a t~aagle of dislocations at another triple-junction 2 may be evidence of the accommoda-
tion of grain-boundary sliding, along one of the boundaries forming that junction and could
be the precursor to fracture at that point.
Figure 25 shows the dislocation structure of a powdei-source berylhium sample which
has seen upset-forged, rolled to a 75-percent reduction at 922 K, and then partly re-
crystallized at 977 K. The structure consists oi a mixture of fine recrystallized grains
in areas of low-oxide particle concentrations and even finer subgrains in areas of high-
oxide concentrations. It should be noted that the central regions of both grains and sub-
grains are relatively free of dislocations. When a similar sample is deformed 10 percent
at room temperature by rolling to simulate flow and fracture during tensile and impact
testing, a high density of dislocations (Fig. 26) are produced in localized areas mainly
around the boundaries of grains or subgrains formed during annealing. This concentra-
tion of dislocations along the grain boundaries is even more apparent in the thicker
regions of the foi (Fig. 27) revealed by the 650-Kv microscope. These dislocation con-
cerntrntior-s in grain-boundary areas are thought to indicate that plastic flow is occurring
preferentially in these regions, probably by grain-boundary sliding leading to grain-
boundary fracture. Two examples of grain-boundary fracture in the very early stager
are shown in Fig. 28. At A , a grain-boundary crack 0. 025 pm wide and 0. 5 lim long
can be seen with difficulty. A larger crack at B can be seen to lie near a grain-boundary
triple ,ur.ction. .Deforme.tion voids 0. 01 pm wide can be seen on oxide particles at A.
The orientation of these voids can be used to determine the stross axis since in compres-
swon the voids are at particle-matrix interfaces at right angles to the compression axis
(Ref. 5). The latter determined in this manner is shown in Fig. 28, amd it can be seen
'hat at leas" one branch of each grain-boundary crack and, in some cases, the heaviest
grain-bound-ry dislocation tangles arc: on boundaries at approximately 45 deg to the
stress axis along the lines of maximum shear stress. ThiB in consistert with the idea
thiat grain-boundary sliding is occurring and Il the precursor to the observed grain-
boundary fractures An indication of the fracture process in coarser,-'grained materia)
24
LOCKHEED PALO ALTO RESEARCH LABORATORV
L 0( t 6 IV # MI1 ,I I 1t 1 01 A CI C 0MP A MV -~ C
A S 0F OC W 0 11 A I NC IA Y COI F IA Y CN
fL1Vsc-1e;:J:38r.j
Fig. 25 '1'ransrm1fjsio.)i E lectron Micrograph (Pt'Iligh-IPurity Plowdvi Source Bei ylliluxnA Ily BollI 32 Ujpset- Forgred, Roiled 75,and Aninealedt :)"I K I1) mmnStructure shows a mnixtuire of fine rcerystallined grains and suhgrains.Magnilication 14,00ox
LOCKHEED PALO ALTrO RESEARCH LABOURATORY
DL0 f i I , A , 0 C A 4 1( ( ( C) IP 0 1 A T
LMSC -D63.336:3
Pi g, 2 6 Transmfission EleIctron Mi crograph of fie .To:( 7' 144) (Boll1 32), Upset-I orged,Rolled 715 ,at 922 K, 11 aurtkilly Recrystallize'd tor 1 hir at 977 K,and Reduced1)'' b"y Cold-Hu(lling. I Ieavy Concentrations of Dislo1(cationls tire seen alongninnly grain hoeund'ar'ies, h ft no cracks have I o ried in this region.Magnification :3!, 000'.:
L:DCKHE ED PALO ALTO Rf.:ESARCH LABO-'RATORY1 )( . .1 1 1 M'. Iý$ýLf 1 $ $p AI(1 ,j . I I IY NC
A uS ,I~ a 1 ~ Ot .. G A1ts , A A Y 0 Tk L
ILMSC -I)633.36(3
PN
Fig. 27 Iligh--Vo!tage (650 kV) Electron Transmission Micrograph (o a lielativelv l'hicktoil of' 1101) 32, 0. 7 :1 Be(), Upset-Forged and Rloiled 75 ,t '92 K, lPartiallyRlecrystdllized at 977 K o)r 5 min and Then Reduced 10'' at 'looml Tolen)era-ture. A. heavy concentration of disiv ation tangles along sonic grain blJ()L1ldAiCscan be seen. Magnification 75,1000'0
37LOCKHEED PALO ALTO RESEARCH L.ABORATOR'Y
•' ~~~A S U S I D? OI A B IY 09 . L0 (K Htt |O A tIS CIAT / (O AIP O S AII A O Y
S 0 .k 4 1 1 . .. ......
Fig. 2~ 8Transm-ission IElectron Micrograph of B~OP :i:-, 7''j Be(), Upset - Forged -.nd Rolled
75"/, at 922 K, Partially Rlecrystallized 977 K I fit-, and D~eformed by Rolling;, i tv atRloom Temperature. A sinall graini--b(undary crack is visible at A, togethier withsomen faintly visible deformation voids on Be() particles which define the compressionaxis. A Larger grain-boundary crack is visible at 13. The cracks and] many oft theioui~daries with a high density of dislocation tangles are a11I)lro~xinlately at 4b) to thecompression axis, i.eC. , aloing the lines ('I1 maximium shear stress. r'VIagnifmc"tiorm20, OOOA' LOCKHEED PALO ALTO RF SEARCH LAB~ORATURY
t ) m I A ( I t ~ C~A ~ . ~ P: IN(38IA % V 6S 1 1A V k O 3F 1i0( Hf I D A IICI A #AO 06,Bp()tA 1I,1
LMSC-1)633363
is shown in Fig. 29. This shows a fully recrystallized (grain size 40 pm) high-purity
ingot-source material (EFI) after 14-percent cold reduction at room temperature.
After a certain amount of plastic flow within the grains, a sharp grain-boundary crack
has been nucleated, which then initiated a second crack, this time on a cleavage plane
within the grain. It can be seen from Fig. 19 that a 40-pmr grain size in EFI fractures
in both an intergranular and a transgranular manner.
5.4 IMPACT RESISTANCE
The impact resistance of as-cast berylliam and dilute beryllium alloys is about 0. 7 J.
This can be increased about 40% by extrusion at 1033 K and partial recrystallization
at 977 K (Fig. 30). However, if the extrusion is then rolled below its recrystalliza-
tion temperature, the impact resistance can be increased to 2.4 J.
Similar results are seen when powder metallurgy extrusions are annealed (Figs. 31 and
32). After ext .sion at 894 K, annealing is required to develop maximum toughness as
shown for BOP 18, while after extrusion at 1089 K sufficient annealing occurs during ex-
trusion to produce maximum toughness in the extruded condition. The effect of beryl-
lium oxide content on the development of maximum toughness during annealing is shown
in Fig. 32. As would be expected, the higher oxide material requires higher tempera-
tures, and its peak toughness is less.
The effect of upset forging prior to rolling at 922 K produces a beneficial tffect oa the
impact strength after annealing as shown in Fig. 33. This is presumably due tu the
greater anisotropy of the deformation and subsequent recrystallization when prior upset
forging is used.
The impact transition temperature of a partially recrystallized powder-source beryl-
lium extrusion, BOP 18, is shown in Fig. 34 to be about 200 K. The impact transition
range of commercially produced cross-rolled sheet is shown for comparison. The
commercial sheet has a poorly defined transition temperature which would appear to
be about 600 K. The transition temperature in hcp metals, including beryllium, has
39
LOCKHEED PALO ALTO RESEARCH LABORATORYL 0 C K H M I 151t kI I a S A CI ( 0 M•P & h Y I N C
A SU I $1 IAIY OY (OC 0 I1lD AIICIAPT COIPO A I I N
Fig, 29 11 aflsmission Elecetron Mi C.r,)griJ)fll of EIA Roloed 96',X, Annealed 977 1•, 1 H ouron 20td) Roled iT.,., A sliv~rp grain I ou da ry C .-r-c k has be en i n ii at d
Ma gulificationl 20 ,)0()()"-
4 0j
LOCKHEE D PALOG ALTO RESEARCH L~a3OfRATORY'1 iIID M % p PA( I ( Qm pA y N (
A t t ~ ý , 0(9NP A I S RA T f I9 J~~
LMSC-633
M C~)
j - H
>LU
_ .2
>L H
7 dd
04 0
00
04-
0
00
0
'.4
(r) 3flVA AdýVHD
41
LOCKHEED PALO ALTO RESEAR~C14 LAB3ORATORYtot WIt is mIlf~il1 a SPA ct COMPANY, INC
A . US S 1,A IV a O taP L *It a1 AIRCIAP! COIPOS 0A IO1P
ILMSC-D633363
I
II
•- 0
o U
- -4
-06
0tý 0Y cd
zm
-- uJ
0
cCti
C-)
0-0 .0
U<LC) '-4
~(r) ]nivA A8dIVHJ
42
S, LOCKHEED PALO ALTO RESEARCH LABORATORY
4OC. IRD I A 51 IO £ I*AoU I OA P AN A NC
A ~~UI44DIA~ ~ C-o 0 L CU II ICS. (IPJ?0
-to a& -- v.-
LMSC -D6 33363
0 ccd
- 04
'00m
0 %0 cv)
X cW LLJ <
0,
0u 2o D u
Ic-4
00
00
/0 <L
04-
(r) gfl1VA AdýYHD
4:
LOCKHEED PALO ALTO RESEARCH LABORAT0,RY
10.41 0~~I SA CI C 0M FA 0 1 1a IJ1 O A av 0F ON 'O~ I 1 A 1 0C IA 17 (OPO0E1A I f
j LMSC-D63336 3
Ib*0
E E
E ~4-4
A EA
0-oz
-0 I-
+ 0
-IJ r z
XX 0 -0 -
LU C,4
LU
(r) If')VA kdSVHDf
44
LOCKHEED PALO ALrO RESEARCH LAB~ORATORY0( .. ,to l if A 14~I * 4A (I (O 0 *M P AkC
A 4tj*IIOAgv Of tO(.A*,9 A.I(VAF? C09O.POA'IO
LMSC -D63:328:3
o Io
- -)
*LUo
p 1- U.
LL LU
LUj
7-1 .t
LIc.-.
-A cnL
0 C lt
(N~(0
4 C'4
LOC~~A~i PAO A~O FSE ACH ABORTOR
4444 4A~i ,¶I~i 4 4 4~(I ( PF -
A 1434 *64 0' K4LNIIO A4*4A' (0SOl~ Oo
LMSC-Dti3:L1(;3
been predicted by A rmstrong (Ref. 7) to be proportional to the square root of the grain
I size. Armstrong shows that it the data of previous investigators who determined tranh--
ition temperature from ductility rather than toughness measurements are plotted in this
J way, the data extrapolate to a transition temperature of 200 K for very fine-grain sizes.
The impact data plotted against d 1/ (Fig. 3,5) break down into three regions which are
differentiated by the amount of dispersoid present. The higher the volume I mriction of
dispersoid, the lower the impact energy at a given grain size. The volume hara' of
intergranular fracture of each sample is also indicated in Fig. 35, and this value will
be used below to estimate the impact resistance ,I fully inl:ergran-alar fracture at each
grain size.
5.5 TLNSILE PROPERTIES
Tensile testing of textured beryllium products can give misleading results since the
strain-to-failure unlike toughness is extremely orientation-dependent, with the maximum
tensile elongation being in the direction of maximum plastic flow during hot working.
For this reason, tersile testing of extrusions was not performed, but an upset forging
of a low-oxide, HIP block was made and tested in tension after variounk aanealijng- treat-
ments. The forging was upset six times in twi directions with the third dijiiension re-
maining unchanged. Since there was zero metal 11ow in thia third direetiun, it s~houl,
be the direction in which tensile elongation is a minimum, an6 the tensile teists were
carried out in this orientation. it has been demonstrated previously (.Ref' 2,4 that, the
texture develope, by upset ferging is less than that normally found in a conventionmt hot-
pressed block and that dhe variation in tensile elongation in an upset forging its 01bout the
same as that in the original billet. It has tns,ý been shown (ReL 2) that the ten•,le
elongation and henri angle of a casting forged ii) only two directions is not greatly depen-
dent on testing direction and is expected to have about the same texture as a billet forged
in three directions.
The mechanical properties of the upset forging are shown In Fig. 36. Compartd to the
''as-fill "! p roperti es, there is an increase, in strength tenfile #.lorlgntti.n, aJnd '111pact
toughness. Th•e peak ductility and' strength occur afteer aniealing, at ]2?00 K, bui the
46
LOCKHEE I' ?',., ALVO ;ESESAFCH LAC8O'NA)FY
I- Ml_ _ -D 1:36
QLQ
0 0(n
CW.4 *6 -jIL
W. - i z
LLJ o ;L 0
0j 0
C.c,
E.c 00C
N cy.- '0 --j
I -J
1- 0 C
'n LLJ PC.) C; C
0 LU 0 O
o0(D * CI
4P4
LOCKHEED PAL,. ALTO RESE~ARCH LABORATORY
A ýU&S1 ,Aq O ko( mfo 1>119.19 CC.19OAATION
I Jmt:c -D63~33(;8
500
ULTIMATE
LZ 400
Lu.
-- j ~0.2%1 YIEL 0Z 300
TESTED) IN DIRECTIONz OF ZE'RO PLASTIC
ui ~~FLOW DURING FRIG
CA ASHP'25KiNIMLJM EXPECTEDTRANS ERSEDUCTILITY
AS--HIr + 1w311 KI RANSVERSE
lAs-ýH1p 1255 KLu LOI'NGITUDINAL
ESTIMATEZ 5
5kAS-HIP +- 1311 KLONGiTUDINAL
ui2A~ I~-
11 L)0 12L0 1300
AS-HiP A1HNEALIN'G TEMPERATUIZe (K)
Fig. 36 TeE.tW~anSeoit Tap npe ~('yo iXaje.f~
I~a~ A I y F ~c~ry. t~ 11 ) B cry,1 1lum P o( Iu cts. T he a mro unt o f L ~granu~irtflou lvaý.-vurc (1.))foi- kiah wiample is shownl
L. OC~U 0 T k,- Rf.ii A~RCH, LASC IRA, ,~ (-,
A - Le t .k f. k pA
1 ,MSC-D633363
hinpact toughness is still increasing at the maximum annealing temperature tested
(1300 K).
5. 6 DISCUSSION
The above results indicate that when th,• grain size of beryllium is reduced below a
certain value, the fracture mode changes fromt cleavage to intergranular. This trans-
ition is inhibited by dispersed particles such as BeO which occur in powder metallurgy
products so that the transition is displaced to finer-grain sizes. It appears from
previous work that even at room temperature grain-boundary sliding precedes inter-
granular fracture and that this additional slip mechanism allows beryllium to satisfy
the Tayloc-Von Mises criterion for homogeneous plastic flow which requires the oper-
ation of five independent slip systems. However, it is apparent that the change of
fracture mechanism alone is only one of the factors contributing to toughneas, since
continued grain refinement of material that fractures in an intergranular manner pro-
duces significant increases in toughness (Fig. 35). Microstructural evidence indicates
that in the toughest materials tested, th.ere is considerable localized flow in grain-
boundary regions before fracture occurs. It is logical to assume, thcrefore, that the
beneficial effect of grain refinement is a result of the enhanced amount of grain-boundary
flow or sliding that can occur before fracture is initiated.
Most of the materials exanmi-ed in this work were only partially recrystallized or in ad-
i vanced stages of recovery, and therefore were fractured only partially in an intergranu-
lar manner. However, it is possible to estimate what the toughness of materials frac-
turing iA a completely intergranular manner at. various grain sizes would be if two as-
sumption8 can be made. The first is that grain refinement has little effect on toughness
if the specimen fractures by cleavage. For example, a high-purity castit)g with grains
10,000 Jim long has the same toughness (0,6b J) when the grain size is reduced to 200 Jim
by upset forging (Ref. 2). It is only when the grain size is further refined and the frac-
ture becomes intergranular that the toughness inwreasos urther support for this view
is provided by L,•ryllium block-tested A-n the as-pressed condition, where oxide particles
S. are present on almost a!l grai.a boundaries and tb.e fracture is over 9i,-percent cleavage.
49
LOCKHEED PALO ALTO RIESCAVRCH LA;ORATI"RY
1 !A 1 tP IC (E I R a ~ I , qP 09
.1 LMSC-D633363
IThese powder-source materials have about the same impact toughness as coarse-
3 grained cast beryllium even though they have a grain size of only 10 Pam.
If the effect of grain refinement on the toughness of beryllium that fractures by cleavage
is negligible and a constant value of 0. 68 J can be assumed, then the improvements in
toughness illustrated in Fig. 34 are a result of the small volume of fine grains that
fracture intergranularly. Since this volume fraction and its grain size are known, it
is possible to calculate what the toughness would be at each grain size if the sample
Sfractured with a 100-percent intergranular fracture. For example, the sample BOP 32
(upset-forged, rolled 75 percent and annealed at 977 K for 10 min) has a measured im-
fpact value of 5.95 J (4.46 ft/lb). Since the fracture is 90-percent cleavage and 10-
percent intergranular, the contribution of the component that fractures by cleavage
will be 0.9 x 0.68 = 0.61 J. The remaining impact resistance contributed by the
material that fractures intergranularly will 3e 5. 95-0. 61 = 5. 34 J. Since, in this
case, this is only 10 percent of the total sample, if the entire sample fractured inter-
granularly at the same grain size, its toughness should be 63.4 J. This calculation
has been carried out for the data shown ir. Fig. 34 and is shown in Fig. 37. It can be
I seen that the projected impact values are considerably hig] er than high-strength alum-
inum alloys such as Al 7075 and at a grain size of 1 j.m approach the toughness of pure
aluminum,
The next questior. is, flow can a uniform fine-grain size of 1 Pm be achieved in beryl-
lium without tihe high-ox!de content formed when conventional powder metallurgy pro-
cesses are used? One answer is that it can be done by the very rapid cooling of beryl-
liumn powder from tie liquid state and consolidating it by techniques that do not allow
•,-ra•n growth. It is hoped that a program along these lines can be initiated in the near
future.
50
LOCKHEEDt PALO ALTO RESEARCH LABORAIORY
I * C I V, 4•) Ylly II K I C e t F T CCM ar9 IAN I1 N
LMSC -D63:3363
I0
04
00 -
oc 0
W LJ.J
S 00 .7
CN.
C7
) -Acv -CD 0 U
-A 93 ^
N C
m uJ
D \l>:) -- W HUJL
N: 0
0~L 0
C) C")
SO (r) AOJNJM IVdVI kASVHD QJIVWNIIS3
51
LOCKHZED PALO ALTO RESEARCH LABORATORY
1 0 U IDII 1 1 LO.CIH..I I A ICEAP 1 O,0A~O
-. ~~ - A A~ 1-------------- 0
Boo
LMSC-D633363
I
Section 6
CONC LUSIONS
(1) The impact toughness of high-purity, low-oxide, powder-source beryllium can be
increased by thermomechanizal treatments to a level not far below that of Al 7075-T6.
(2) Optimum toughness occurs when the microstructure consists of very fine (< 4 Am)
recrystallized grains or subgrains which are on the threshold of recrystallizing in
situ, and which fracture intergranularly.
(3) There is a grain-size controlled fracture mode change in both power-source and
ingot-source beryllium. The change is ,]isplaced to finer grain sizes by oxide
particles and texture.
(4) There is potential for a considerable further improvermnt in the toughness of beryl-
lium by the use of rapidly cooled powders.
(5) To produce the finest grain size in beryllium for a given total reduction, a thermo--
mechanical process should avoid any recrystallization before the final annealing
treatment.
(6) RecrystallIzation in heavily deformed beryllium occurs by the in situ transforma-
tion of subgrains to grains by the dislocation migration from grain interiors to
grain boundaries.
(7) Upset forging followed by annealing can produce increases in strength, ductility,
and toughness.
52
LOCKHEED PAiLO ALTO RESEARCH L.AdORATOR'"L 0 C&I N I 1I A MI V S I LII I A C 9 A C COMVAP. , IV l I C
A $,U I•J SI I A Oa L 0 C IN I1 0 A - 9 4 II C• 11 1 F 0 A T ION
AL.
S~LMSC-D633363
II
f Section 7
REFERENCES
1. D. Webster, R. L. Greene, R. W. Lawley, and G. J. London, "Factors Affecting
the Tensile Strength, Elongation and Impact Resistance of Low Oxide, Hot Isostati-
cally Pressed Beryllium Block, Met. Trans., Vol. 7A, Jun 1976, p. 851
2. D. Webster and D. D. Crooks, Improved Beryllium Ductility Study, Contract
NC,6921-74-C-0114, Final Report to Naval Surface Weapons Center, LMSC-D507268,
Aug 1976
3. V. E. Ivanov, G. F. Tikhinskij, I. I. Papirov, I. A. Taranenko, E. S. Karpov,
arid ti. S. Kapcherin, "Plastic and Superplastic Deformation of Fine Grained High
Purity Beryllium," Proceedings of Fourth International Conference on Beryllium,
London, 1977
4. D. Webster, R. L Greene, and R. W. Lawley, "Factors Controlling the Strength
and Ductility of High Purity Beryllium Block," Met. Trans., Vol. 5, Jan 1974,
p. 91
5. D. Webster, "Grain Growth and Recrystailization in Thoric Dispersed Nickel and
Nichrome," TrMS. AIME, Vol. 242, Apr 1968, p. 640
6. Lock"heed Missiles & Space Co., inc., Final Report for Evaluation of Beryium
for Space Shuttle Coinponents, LMSC-D159319, Contract NAS 8-27739, Sunnyvale,
California, 1972
7. R. W. Armstrong, Acta Met. Vol. 1.6, 1968, p. 347
53
j LOCKHE, L PALO ALTO RESEARCH LABORATORY
1 0 C K M I I D M I S I Il $ a I P A C I C 0 04P A r I A II
A SU5S1U0IIA I 1 Ot L0C KM I 1 A II C IA FI C C3 I 01A t 1 0